Comparing Inaturalist and Traditional Surveys for Estimating Richness, Diversity, and Abundance of Reptiles and Amphibians

Authors:
Open Access
Publisher website
Google Scholar
Abstract
Community science (CS) data can increasingly address a range of interesting questions in herpetology. However, data from unstructured CS projects like iNaturalist are collected in a very different manner than in traditional herpetological surveys, and this could lead to erroneous conclusions about herpetological communities. We compared inferences about species richness, diversity, and the representation of individual species from 71 sites in the United States that had both published herpetological surveys and adequate samples of iNaturalist observations. After adjusting for sample size, we found no significant differences in observed species richness between traditional survey data and iNaturalist data, and we found relatively high correlations between the two data sources in observed species richness across sites. However, the overlap between species lists derived from iNaturalist versus traditional surveys was only moderate, with reptile lists typically having 40–60% shared species and amphibian lists having 50–75% shared species. Furthermore, species diversity, and particularly evenness, differed consistently between the data sources, with observations more evenly spread among species in iNaturalist data as compared to traditional survey data. Likewise, correlations in species counts between the two data sets at individual sites were often low, and correlations in species diversity across sites were much lower than correlations in species richness. With respect to individual species, some species in each taxonomic group were consistently over- and under-represented in iNaturalist data compared to traditional survey data. Over-represented species included Red-backed Salamanders, Fowler's Toads, Eastern Fence Lizards, and Gopher Snakes, whereas under-represented species included White-spotted Slimy Salamanders, Gray Treefrogs, Six-lined Racerunners, Milk Snakes, and Musk Turtles. Overall, our results suggest that unstructured CS data could be useful for estimating species richness at individual sites or for comparing species richness across sites, particularly when sample sizes are high (e.g., hundreds to thousands of observations). However, these data sources should not be used to analyze community structure or diversity, at least without correcting for biases in CS data with respect to how likely different species are to be recorded.

This publication has 41 references indexed in Scilit: